(++)
If participants are unaware they are being observed, they will behave naturally. E.g, no demand characteristics.
(-)
Observer bias. It is easy to see what you want to see.
(–)
It may be difficult to ensure inter-rather reliability (the extent to which observers agree on observational recordings).
Ecological Validity.
The extent to which a study resembles true life.
(+)
If the behaviour is o served in a natural setting, then there will be high ecological validity.